I've checked in some changes to the libdfp branch. Notable changes include: moving the conditional overrides to the base files, adding an --enable-decimal-float option option to the configure script, and changing the makefile to build a single libdfp.so and no longer build libdecnumber.so. The GCC trunk 4.3.0 version currently has libdecnumber symbol collisions with libdecnumber in libdfp.so, but there should be a GCC patch fix forthcoming. GCC 4.2.1 works fine since it uses an older version of decnumber.
Changelog:
* stdlib/stdlib.h: Merged DFP code.
* wcsmbs/wchar.h: Merged DFP code.
* configure.in: Added --enable-decimal-float option.
* math/math.h: Merged DFP code.
* math/fenv.h: Merged DFP code.
* math/bits/mathcalls.h: Merged DFP code.
* config.h.in: Added __STDC_DEC_FP__ define.
* stdio-common/printf.h: Merged DFP code.
* stdio-common/vfprintf.c: Merged DFP code.
* stdio-common/vfscanf.c: Merged DFP code.
* stdio-common/printf-parse.h: Merged DFP code.
* configure: Regenerated.
* dfp/sysdeps/powerpc/powerpc32/power6/fpu/Versions: Changed to GLIBC_2.6.
* dfp/sysdeps/powerpc/powerpc64/power6/fpu/Versions: Changed to GLIBC_2.6.
* dfp/sysdeps/soft-dfp/Versions: Changed to GLIBC_2.6.
* dfp/sysdeps/dfp/stdlib/stdlib.h: Deleted file.
* dfp/sysdeps/dfp/stdio-common/vfprintf.c: Deleted file.
* dfp/sysdeps/dfp/stdio-common/printf-parse.h: Deleted file.
* dfp/sysdeps/dfp/stdio-common/vfscanf.c: Deleted file.
* dfp/sysdeps/dfp/stdio-common/printf.h: Deleted file.
* dfp/sysdeps/dfp/Versions: Changed to GLIBC_2.6
* dfp/sysdeps/dfp/math/math.h: Deleted file.
* dfp/sysdeps/dfp/math/bits: Deleted file.
* dfp/sysdeps/dfp/math/bits/mathcalls.h: Deleted file.
* dfp/sysdeps/dfp/math/fenv.h: Deleted file.
* dfp/sysdeps/dfp/wcsmbs/wchar.h: Deleted file.
* dfp/Versions.def: Changed to GLIBC_2.6.
* dfp/Versions: Changed to GLIBC_2.6.
* dfp/Makefile: No longer build libdecnumber.so, instead only libdfp.so.
--
Pete Eberlein
IBM Linux Technology Center
Linux on Power Toolchain